Slow conformational dynamics of the guanine nucleotide-binding protein Ras complexed with the GTP analogue GTPcS

Transcription

1 Slow conformational dynamics of the guanine nucleotide-binding protein Ras complexed with the GTP analogue GTPcS Michael Spoerner 1, Andrea Nuehs 1, Christian Herrmann 2, Guido Steiner 1 and Hans Robert Kalbitzer 1 1 Universität Regensburg, Institut für Biophysik und physikalische Biochemie, Germany 2 Ruhr Universität Bochum, Physikalische Chemie I, Germany Keywords conformational equilibria; GTP analog; GTPcS; Ras Correspondence H. R. Kalbitzer, Institut für Biophysik und physikalische Biochemie, Universitätsstraße 31, Regensburg, D-93040, Germany Fax: Tel: hans-robert.kalbitzer@biologie. uni-regensburg.de (Received 28 July 2006, revised 13 November 2006, accepted 8 January 2007) doi: /j x The guanine nucleotide-binding protein Ras occurs in solution in two different conformational states, state 1 and state 2 with an equilibrium constant K 12 of 2.0, when the GTP analogue guanosine-5 -(b,c-imido)triphosphate or guanosine-5 -(b,c-methyleno)triphosphate is bound to the active centre. State 2 is assumed to represent a strong binding state for effectors with a conformation similar to that found for Ras complexed to effectors. In the other state (state 1), the switch regions of Ras are most probably dynamically disordered. Ras variants that exist predominantly in state 1 show a drastically reduced affinity to effectors. In contrast, Ras(wt) bound to the GTP analogue guanosine-5 -O-(3-thiotriphosphate) (GTPcS) leads to 31 P NMR spectra that indicate the prevalence of only one conformational state with K 12 > 10. Titration with the Ras-binding domain of Raf-kinase (Raf-RBD) shows that this state corresponds to effector binding state 2. In the GTPcS complex of the effector loop mutants Ras(T35S) and Ras(T35A) two conformational states different to state 2 are detected, which interconvert over a millisecond time scale. Binding studies with Raf- RBD suggest that both mutants exist mainly in low-affinity states 1a and 1b. From line-shape analysis of the spectra measured at various temperatures an activation energy DH 1a1b of 61 kjæmol )1 and an activation entropy DS 1a1b of 65 JÆK )1 Æmol )1 are derived. Isothermal titration calorimetry on Ras bound to the different GTP-analogues shows that the effective affinity K A for the Raf-RBD to Ras(T35S) is reduced by a factor of about 20 compared to the wild-type with the strongest reduction observed for the GTPcS complex. Guanine nucleotide-binding proteins of the Ras superfamily function as molecular switches, cycling between a GDP-bound off and a GTP-bound on state. They regulate a diverse array of signal transduction and transport processes. It has been shown using 31 P NMR spectroscopy that Ras (rat sarcoma) protein occurs in two conformational states (state 1 and 2) when complexed with the GTP analogues guanosine-5 -(b,c-imido)triphosphate (GppNHp) [1] or guanosine-5 -(b,c-methyleno)triphosphate (GppCH 2 p) [2]. These two states interconvert with rate constants in the millisecond time scale. They are characterized by typical 31 P NMR chemical shifts, with shift differences up to 0.7 p.p.m. NMR structural studies have shown that this dynamic equilibrium comprises two regions of Abbreviations GppCH 2 p, guanosine-5 -(b,c-methyleno)triphosphate; GppNHp, guanosine-5 -(b,c-imido)triphosphate; GTPcS, guanosine-5 -O-(3- thiotriphosphate); ITC, isothermal titration calorimetry; Raf-RBD, Ras-binding domain of Raf-kinase; Ras, protein product of the proto oncogene ras (rat sarcoma). FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS 1419

2 Conformational dynamics of Ras bound to GTPcS M. Spoerner et al. the protein called switch I and switch II [1,3,4]. Solidstate NMR shows that even in single crystals or crystal powders of Ras(wt) Mg 2+ GppNHp the two conformational states can be observed to be in dynamic equilibrium at ambient temperatures [5,6]. A threonine residue located in the effector loop (Thr35 in Ras) is conserved in all members of the Ras superfamily and seems to play a pivotal role in the conformational equilibrium. It is involved, via its side-chain hydroxyl, in the coordination of the divalent metal ion and, via its main-chain amide, in a hydrogen bond with the c-phosphate of the nucleotide when complexed to the effector [7,8]. The same coordination pattern is most probably preserved in state 2 of free Ras. Replacing this threonine in Ras with an alanine or serine residue leads to a complete shift of the equilibrium towards state 1 in solution, when Ras is bound to the GTP analogues GppNHp [9] or GppCH 2 p [2]. These Ras variants, previously used as partial loss-of-function mutants in cell-based assays, show a reduced affinity between Ras and effector proteins without Thr35 being involved in any interaction. X-Ray crystallography [9] on Ras(T35S) Mg 2+ GppNHp and EPR investigations [10] suggest that switch I and switch II exhibit high mobility in state 1. Recently, X-ray structures of M-Ras [11] and of the G60A mutant of human H-Ras [12], both in the GppNHp-bound form, were published. These Ras variants seem to exist in conformational state 1, as shown using 31 P NMR spectroscopy. In the X-ray structure the contacts of Thr35 (Thr45 in M-Ras) with the metal ion and the c-phosphate group do not exist. 31 P NMR data indicate that state 2 corresponds to the conformation of Ras found in complex with the effectors. State 1, characteristic of the mutants Ras(T35S) and Ras(T35A) in the GppNHp form, represents a weakbinding state of the protein [9,13]. Upon addition of the Ras effector Raf-kinase, the 31 P NMR lines of Ras(T35S) but not Ras(T35A) shift to positions corresponding to the strong binding conformation of the protein [9]. A conformational equilibrium in the interaction site with effectors seems to be a general property of Ras and other small GTPases [14]. The equilibrium is influenced not only by specific mutations but also by the nature of the GTP analogue bound (GppNHp or GppCH 2 p). In this study we investigate the dynamic behaviour of Ras in complex with guanosine-5 -O-(3- thiotriphosphate) (GTPcS), another commonly used GTP analogue that is hydrolysed slowly to find more evidence for the biological importance of the conformational equilibria. Results Chemical shifts of the nucleotide analogue GTPcS in the absence and in the presence of magnesium ions Chemical shift values for the phosphates and the thiophosphate group of the nucleotide depend strongly on the degree of protonation of their oxygens. Furthermore, chemical shifts and pk values are influenced by Mg 2+ binding to the protein nucleotide complex. For a better interpretation of the chemical shifts of the protein-bound nucleotide analogue we first studied GTPcS in the presence and absence of Mg 2+ ions within a ph range of The rate of exchange between Mg 2+ and the nucleoside triphosphate is slow enough to observe the resonances of the metal-free form separately from the metal-complexed form at lower temperatures. Therefore, experiments were performed at 278 K to ensure that over the whole ph range a significant contribution of metal-free nucleotide, if existing, could be directly detected by additional resonance lines. At a magnesium concentration of 3 mm the nucleotide is completely saturated with the divalent ion in the ph range studied since further increase of the Mg 2+ concentration does not influence the observed chemical shifts (also see Experimental procedures). Figure 1 shows the titration curves for GTPcS in the absence and presence of Mg 2+. Separation of the three phosphate signals by more than 60 p.p.m. is rather large. Particularly in case of the c-phosphorus (Fig. 1A,B) two pk values are necessary in order to describe the observed dependence of chemical shifts in the ph range studied. The corresponding pk values and chemical shifts are summarized in Table 1 together with the data for the analogues GppNHp and GppCH 2 p [2]. As expected, the apparent pk values decrease substantially in the presence of the metal ion. By far the largest effect on the chemical shifts is found for the b- and c-phosphate group, but a slight shift of 0.6 p.p.m. is also seen for the a-phosphorus resonance in the Mg 2+ GTPcS complex. In agreement with previous studies on ATP [15], our data suggest a mixture of different metal complexes in solution with a high population of complexes where the b- and c-phosphate is involved, as shown previously for the GTP analogues GppNHp and GppCH 2 p [2]. The pk 3 values in GTPcS are much smaller than those reported for GppNHp and GppCH 2 p. The value of pk 2 does not depend much on the analogue when a relatively large error is taken into consideration. pk 2 and pk 3 are usually associated with the first and the second 1420 FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS

3 M. Spoerner et al. Conformational dynamics of Ras bound to GTPcS Fig. 1. Titration curves of free and Mg 2+ bound GTPcS. (A,C) 31 P chemical shift values of the a-, b- and c-phosphate groups were determined on a 2.5 ml of a 1 mm GTPcS solution in 100 mm Tris, 95% H 2 O and 5% D 2 O containing 0.1 mm 2,2-dimethyl-2-silapentane-5-sulfonate for indirect referencing. The ph was adjusted by adding HCl or NaOH. Measurements were performed in a 10-mm sample tube at 278 K. (B,D) Measurements on the Mg 2+ complexes were performed in the presence of 3 mm MgCl 2. The dependence of chemical shifts on the ph values was fitted to Eqn (7). The 31 P resonances were assigned by selective 1 H- and 31 P-decoupling experiments. Table 1. ph dependence of chemical shifts of different GTP analogues. Data were recorded at 278 K in solutions of 1 mm nucleotide in the absence or presence of 3 mm MgCl 2 in 95% H 2 O 5% D 2 O. In a first approximation d 2, d 3, and d 4 correspond to the chemical shifts of two-, three-, and fourfold negatively charged nucleotide. pk 2 and pk 3 are the corresponding pk a values of the three phosphates of the nucleotide. d 2 values are given in parentheses the titration up to ph 1.5 does not allow the precise estimation of this value. For d 3 and d 4 the estimated error is ± 0.05 p.p.m. Nucleotide Phosphate group d 2 p.p.m. pk 2 d 3 p.p.m. pk 3 d 4 p.p.m. GTPcS a ()11.3) ) ) b ()24.0) 2.8 ± 0.1 ) ± 0.02 ) c (40.8) Mg 2+ GTPcS a ()11.2) ) ) b ()24.2) 1.7 ± 0.5 ) ) c (41.6) ± GppCH 2 p a a ()10.86) ) ) b (7.14) 3.2 ± ± c (17.85) ± Mg GppCH 2 p a a ()10.83) ) ) b (9.50) 2.3 ± c (16.98) GppNHp a a ()10.95) ) ± 0.02 ) b ()12.27) 3.4 ± 0.04 ) ) 7.76 c (0.20) ) 1.64 ) 0.91 Mg GppNHp a a ()11.17) ) ± 0.02 ) b ()9.36) 2.0 ± 0.8 ) 8.95 ) 5.46 c ()1.38) ) 2.16 ) 1.02 a Data from Spoerner et al. [2]. deprotonation step at the c-phosphate group of the nucleotide for the transition from the threefold negatively charged state to the fourfold negatively charged state. In line with this suggestion the largest shifts are observed for the c-phosphate group for the first deprotonation step for the three analogues. However, the second deprotonation step is associated with larger changes in the b-phosphate shifts in GppNHp and GppCH 2 p, indicating a more complex ph perturbation of the electronic system in these analogues. FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS 1421

4 Conformational dynamics of Ras bound to GTPcS M. Spoerner et al. Conformational states of Ras complexed with Mg 2+ GTPcS Figure 2 shows 31 P NMR spectra of Ras(wt) in complex with the slowly hydrolysable GTP analogue GTPcS at various temperatures. Assignment of the resonance lines was confirmed by a 2D 31 P 31 P NOESY experiment on Ras(wt) Mg 2+ GTPcS (data not shown). Binding of GTPcS to the Ras protein leads to rather large chemical shift changes. In contrast to the observations made for the GTP analogues GppNHp and GppCH 2 p [1,2] only one set of resonances can be observed for the wild-type protein in the temperature range K (Fig. 2). This most probably means that wild-type Ras occurs predominantly in one state when GTPcS is bound. It is reasonable to assume that a second structural state also exists and is characterized by different chemical shift values, as observed in the GppNHp and GppCH 2 p complexes [1,2]. When this second state has clearly different chemical shifts compared with the first state then two scenarios are consistent with the observed spectrum. If fast exchange conditions prevail over the whole temperature range, then only one averaged resonance signal per phosphate group would be observed. If slow exchange conditions prevail, a second conformational state, characterized by clearly different chemical shifts, must have a rather low population because no signals can be detected above noise level. In this case, from the signal-to-noise ratio the equilibrium constant for the two states can Fig P NMR spectra of wild-type Ras complexed with Mg 2+ GTPcS at various temperatures. The samples contained 1mM Ras(wt) Mg 2+ GTPcS in 40 mm Hepes NaOH ph 7.4, 10 mm MgCl 2, 150 mm NaCl, 2 mm 1,4-dithioerythritol and 0.1 mm 2,2-dimethyl-2-silapentane-5-sulfonate in 5% D 2 O, 95% H 2 O, respectively. The absolute temperature was controlled by immersing a capillary with ethylene glycol and measuring the hydroxylmethylene shift difference [28]. be estimated to be > 10. Analysing the temperature dependence of the line width, particularly of the c-phosphorus resonance, slow exchange conditions are more likely. At lower temperatures the line width decreases with increasing temperature due to the decrease of the rotational correlation time. At higher temperatures the line width increases again (51 Hz at 298 K, 57 Hz at 303 K). Chemical shift also changes within the temperature range of K by p.p.m. At higher temperatures, the GTP analogue hydrolyses, and resonances of Ras-bound GDP are thus detected. In principle, one would expect to observe thiophosphate and Ras-bound GDP as result of GTPcS hydrolysis. In contrast, with all the measurements performed in this study, inorganic phosphate could be observed only using 31 P NMR. In addition, H 2 S could be detected by its smell after a time. The exact mechanism of thio phosphate decay could not be clarified. It is dependent on the presence of Ras, but may be also due to other protein impurities occurring in low concentrations in the Ras preparations. In contrast to the situation observed for wild-type protein in the complex of GTPcS with the mutant Ras(T35S) or Ras(T35A), additional 31 P NMR lines are found at low temperature (Fig. 3A). With increasing temperature, the lines initially become broader before coalescing again at higher temperature (Fig. 4A). From our studies with GppNHp and GppCH 2 p we expect that the effector interaction state 2 becomes destabilized by replacing Thr35 with a serine or an alanine residue, and therefore at least one of the new lines seen in the mutant is likely to correspond to state 1. Because no component of the two sets of resonances of Ras(T35S) and Ras(T35A) has a chemical shift that corresponds to that of Ras(wt) it is not clear whether the two sets of resonance lines correspond to state 1 and state 2 or if they represent two substates of state 1 (see below). In the following, we call them state 1a and state 1b. The equilibrium constant K 1a1b ¼ [1b] [1a] between these two states is 0.5. In the case of the serine mutant, a weak third line of the c-phosphorus signal with a similar chemical shift to the resonance of wild-type Ras seems to exist (Fig. 3A); this is not visible in the spectrum of the T35A mutant. The chemical shifts are summarized in Table 2. With knowledge of the resonance positions corresponding to state 1a and 1b, we investigated whether these states also exist in wild-type Ras bound to GTPcS. Separation of the chemical shift values between state 1b and state 2 of more than 4 p.p.m. allowed us to perform a saturation-transfer experiment with presaturation at frequencies around the signal corresponding to state 1b. If exchange occurs over a 1422 FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS

5 M. Spoerner et al. Conformational dynamics of Ras bound to GTPcS A B Fig. 3. Conformational equilibria of wild-type Ras and Ras mutants complexed with Mg 2+ GTPcS. (A) The sample contained 1 mm Ras(wt) Mg 2+ GTPcS (lower), 1.2 mm Ras(T35S) Mg 2+ GTPcS (middle), and 1 mm Ras(T35A) Mg 2+ GTPcS (upper) in 40 mm Hepes NaOH ph 7.4, 10 mm MgCl 2, 150 mm NaCl, 2 mm 1,4-dithioerythritol and 0.1 mm 2,2-dimethyl-2-silapentane-5-sulfonate in 5% D 2 O, 95% H 2 O, respectively. Data were recorded at 278 K. The assignment was determined by a 31 P 31 P NOESY experiment on Ras(wt) Mg 2+ GTPcS. 31 P resonances assigned to Ras nucleotide complex in conformation of state 1a or state 1b are coloured in red, the resonances assigned to state 2 are coloured green. (B) 31 P NMR saturation transfer experiment on Ras(wt) Mg 2+ GTPcS. The integrals of the resonance corresponding to the c-thiophosphate group in state 2 of Ras(wt) are given in dependence of the frequency of presaturation d. For presaturation a weak rectangular pulse of 1 s duration and a B 1 -field of 18 Hz were used. A Lorentzian function was fitted to the data. The integral of the c-phosphorus signal without presaturation is set to 100%. Fig. 4. Experimental and simulated 31 P NMR data of Ras(T35S) Mg 2+ GTPcS at different temperatures. The sample contained 1.2 mm Ras Mg 2+ GTPcS in40mm Hepes NaOH ph 7.4, 10 mm MgCl 2,2mM 1,4-dithioerythritol and 0.1 mm 2,2-dimethyl-2-silapentane-5-sulfonate in 5% D 2 O, 95% H 2 O. The absolute temperature was controlled by immersing a capillary with ethylene glycol and measuring the hydroxyl methylene shift difference [28]. (A) Experimental spectra; (B) simulated spectra. Experimental data were filtered by an exponential filter leading to an additional line broadening of 5 Hz. Total number of scans per spectrum were The rate constant for the transition state 1a to state 1b are indicated. Data were simulated as described in Experimental procedures. The transverse relaxation rates 1 T 2 at 278 K (in the absence of exchange) obtained from the data analysis are 251 s )1 for both state 1a and state 1b of the a-phosphate group of bound GTPcS, 236 s )1 and 204 s )1 for the b-phosphate group of bound GTPcS in state 1a and state 1b, respectively, and 189 s )1 for state 1a and 1b of the bound c-thiophosphate group (values are given with an estimated error of ± 15 s )1 ). timescale < T 1 a decrease in the integral of the resonance corresponding to state 2 should be observed, even when state 1 is too sparsely populated to be detectable directly. Some results are shown in Fig. 3B. A minimum of the resonance integral of state 2 is obtained at a presaturation frequency of 32.7 p.p.m., which corres- FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS 1423

6 Conformational dynamics of Ras bound to GTPcS M. Spoerner et al. Table P chemical shifts and conformational states of Ras complexed with different GTP analogues. Data were recorded at various temperatures. Shifts were taken from spectra recorded at 278 K. The equilibrium constant K 12 between state 1 and 2 is calculated from integrals of the c-thiophosphate resonances defined by K 12 ¼ k 12 k 21 ¼ [2]] ([1a] + [1b]). State 2 is assigned to the conformation close to the effector binding state. The error is < 0.03 p.p.m. for the chemical shifts and < 0.1 for the equilibrium constants. ND, not detected. a-phosphate b-phosphate c-phosphate Ras-complex d 1 (p.p.m.) d 2 (p.p.m.) d 1 (p.p.m.) d 2 (p.p.m.) d 1 (p.p.m.) d 2 (p.p.m.) K 12 K 1a1b a Ras(wt) Mg 2+ GTPcS )11.30 ) > 10 ND b Ras(T35S) Mg 2+ GTPcS )10.70 )17.96 a )17.22 a a Ras(T35A) Mg 2+ GTPcS )10.80 )17.92 a a < )17.19 a a Ras(wt) Mg 2+ GTPcS )11.19 ) > 10 ND b + Raf-RBD Ras(T35S) Mg 2+ GTPcS )11.22 ) > 10 ND b + Raf-RBD Ras(T35A) Mg 2+ GTPcS )10.50 ) a < Raf-RBD a a Chemical shifts in state 1a (lower) and 1b (upper). b Values could not be determined since signal cannot be detected. ponds to the frequency of state 1b detected for the two Thr35 mutants. These results indicate the existence of state 1b in wild-type Ras, but with a very sparse population. A more detailed analysis including calculation of exchange rates was not possible because of the limited signal-to-noise. Dynamics of the conformational exchange By analysing the temperature dependence of the 31 P NMR data from Ras(T35S) Mg 2+ GTPcS (Fig. 4B) for the transition between substates 1a and 1b the Gibb s free activation energy DG, the activation enthalpy DH and the activation entropy DS can be determined (Table 3) using a full-density matrix analysis. The exchange rates obtained are somewhat higher than that found between states 1 and 2 of Ras(wt) Mg 2+ GppNHp or Ras(wt) Mg 2+ GppCH 2 p. Whereas DG of the exchange in Ras(T35S) Mg 2+ GTPcS is equal to that obtained for the other complexes, both DH, and DS are somewhat lower. For the other nucleotides studied, relaxation times T 2 at 278 K for the a- and c-phosphate group were quite different for the two conformational states 1 and 2. We did not find such large differences between the corresponding T 2 relaxation times for the conformational states 1a and 1b of Ras(T35S) Mg 2+ GTPcS. Complex of Ras Mg 2+ GTPcS with the Ras-binding domain of Raf-kinase Addition of the Ras-binding domain of Raf-kinase (Raf-RBD) to Ras(wt) Mg 2+ GTPcS leads to line broadening of the resonances (Fig. 5, Table 2), but only to very small changes in the chemical shifts ( Dd 0.16 p.p.m). This is in line with the assumption that the wild-type protein occurs mainly in conformational state 2 when the GTP analogue GTPcS is bound. Correspondingly, in Ras(T35S) Mg 2+ GTPcS, lines preliminary assigned to states 1a and 1b decrease in intensity when Raf-RBD is bound, whereas the intensity of lines located close to those assigned in wild-type Ras to state 2 increases (Fig. 5, Table 2). The changes in chemical shift induced by Raf binding are rather large in the mutant, suggesting that none of the states visible in the spectrum of Ras(T35S) Mg 2+ GTPcS corresponds to state 2 found in the wild-type protein. Complex formation between Raf-RBD and Ras(T35A) Mg 2+ GTPcS (Fig. 5, Table 2) leads only to a line broadening of the two lines of the c-phosphate group, and not to significant changes in chemical shift or the relative populations of the resonances. In particular, the relative intensity of the downfield-shifted c-phosphorus resonance is not increased in the presence of the effector as would be expected if it corresponded to effector binding state 2. Influence of the GTP analogue on the affinity between Raf-RBD and Ras The affinities of wild-type and (T35S)Ras complexed with the different GTP analogues GppNHp, GppCH 2 p and GTPcS to Raf-RBD were determined using isothermal titration calorimetry (ITC) at 298 K in a buffer identical to that used in the NMR spectroscopy experiments. Within the limits of error, the effective 1424 FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS

7 M. Spoerner et al. Conformational dynamics of Ras bound to GTPcS Table 3. Exchange rates and thermodynamic parameters in different Ras nucleotide complexes. The rate constants k 12 and k 21 (k 1a1b and k 1b1a ) were calculated by a line-shape analysis based on the density matrix formalism as described in Experimental procedures. The free activation energy DG, the activation enthalpies DH, and the activation entropies DS, were calculated from the temperature dependence of the exchange rates on the basis of the Eyring equation. The values for the transition between state 1 and state 2 k 12 and k 21 are given. The states are defined as in Table 1. DG 12 or DG 1a1b is the difference in free enthalpy between state 2 (1b) and 1 (1a). T 2 times given are without exchange contribution and were obtained from the line shape analysis. The estimated error is ± 0.3 ms. Protein complex Temp. (K) Exchange rate constant (s )1 )_ DG j 1a1b DH j 1a1b TDS j 1a1b DG 1a1b k 1a1b k 1b1a (kjæmol )1 ) (kjæmol )1 ) (kjæmol )1 ) (kjæmol )1 ) Ras(T35S) Mg 2+ GTPcS ± 2 61 ± 1 18 ± ± 0.15 a k 12 k 21 DG j 12 DH j 12 TDS j 12 DG 12 Ras(wt) Mg 2+ GppNHp a ± 5 70 ± 3 28 ± 2 ) 1.48 ± Ras(wt) Mg 2+ GppCH 2 p a ± 5 63 ± 3 29 ± 2 ) 1.65 ± Relaxation times T 2 (ms) of the resonances of Protein-complex a-phosphate b-phosphate c-phosphate (1) or (1a) (2) or (1b) (1) or (1a) (2) or (1b) (1) or (1a) (2) or (1b) Ras(T35S) Mg 2+ GTPcS Ras(wt) Mg 2+ GppNHp a Ras(wt) Mg 2+ GppCH 2 p a a Data from Spoerner et al. [2]. Note that the values given differ somewhat from those given by Geyer et al. [1] because absolute temperature was controlled independently and the new assignment of the signals were considered. association constant K A between wild-type Ras and Raf-RBD is not influenced by the type of bound analogue (Table 4). However, in all cases, the contributions of enthalpy and entropy to DG differ between nucleotide analogues. Although for the Thr35 mutant the error ranges for the three nucleotide analogues overlap, a difference in affinities between Ras(T35S) bound to the analogue GTPcS, where the oxygen between b- and c- phosphate is still available, and GppCH 2 p may exist. A significant decrease in K A, by a factor of 20, is seen, independent of the analogue used when the wild-type protein is compared with Ras(T35S). The decrease in affinity is due to changes in DH and DS, which partly compensate. Discussion The environment of the nucleotide bound to the protein NMR spectroscopy very sensitively reports changes in the environment of a given atom by measuring a change in its resonance frequency. Whenever chemical shift changes are visible they indicate that there is a change in the environment of the observed nucleus. For phosphorus resonance spectroscopy on nucleotides, it is known that two factors mainly determine chemical shift changes, a conformational strain and electric field effects polarizing the oxygen atoms of the phosphate groups. In addition to these direct effects, long-range effects may occur that are caused by a structure-dependent change in the anisotropy of the magnetic susceptibility. Here, ring current effects may be the most dominant contribution. We have previously studied the complexes of Ras using the GTP analogues GppCH 2 p and GppNHp [2], which differ in the position of the b c-bridging oxygen by replacing the naturally occurring oxygen either with an apolar group or a hydrogen-bond donator. We have now completed the picture using the slowly hydrolysing GTP analogue GTPcS, in which the b cbridging oxygen is not affected, but the physicochemical properties of the c-phosphate group are modified. For a quantitative analysis of the chemical shift changes induced by protein binding it was necessary to have reliable data for the system not perturbed by FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS 1425

8 Conformational dynamics of Ras bound to GTPcS M. Spoerner et al. Fig p NMR spectra of wild-type Ras and Ras mutants bound to Mg 2+ GTPcS in complex with Raf-RBD. Initially the samples contained 1.0 mm Ras Mg 2+ GTPcS (lower), 1.2 mm Ras(T35S) Mg 2+ GTPcS (middle) or 1.0 mm Ras(T35A) Mg 2+ GTPcS (upper) in 40 mm Hepes NaOH ph 7.4, 10 mm MgCl 2, 150 mm NaCl, 2 mm 1,4-dithioerythritol and 0.1 mm 2,2-dimethyl-2-silapentane-5-sulfonate in 5% D 2 O, 95% H 2 O, respectively. A solution of 9.8 mm Raf-RBD dissolved in the same buffer was added in increasing amounts. The molar ratios of Raf-RBD Ras are 1.5 for Ras(wt) and 2 in the mutant samples. Data were recorded at 278 K. 31 P resonances assigned to Ras nucleotide complex in conformation of state 1a or state 1b are coloured red, the resonances assigned to state 2 are coloured green. protein binding that we provide here. Although data had been published previously for free GTPcS [16], they were measured under different experimental conditions and the referencing system (external standard) in particular is not sufficiently reliable for precise comparisons. When one compares the chemical shift changes Dd in the free Mg 2+ nucleotide complexes (Table 1) with those induced by protein binding (Table 2) one may obtain information on the change of the environment of the phosphate groups in the different complexes. In wild-type Ras in state 2, one finds Dd values of )0.26, 6.39 and 3.10 p.p.m., respectively for the a-, b- and c-phosphate of GTPcS. The corresponding shift changes are )1.15, 7.51 and )2.41 p.p.m. for GppNHp and )2.44, 6.32 and )3.03 p.p.m. for GppCH 2 p. The a-phosphate groups in the three GTP analogues should be least influenced by the modifications. In accordance with this observation, in the absence of protein, their response to a change in ph (acidity) is very small, only an upfield shift of < 0.26 p.p.m. is observed when the c-phosphate group is protonated by a decrease in ph. After binding to the protein, for all three analogues an upfield shift between of 0.26 and 2.44 p.p.m. is observed, indicating that the environmental changes are qualitatively similar but differ in detail. Potential phosphate group interactions can be derived from the published X-ray structures, although one should be aware that they show differences in effector loop details that may reflect the occurrence of different conformational states in solution. Because NMR data indicate that the interaction of Ras with Raf-RBD stabilizes the effector loop in a welldefined, state 2-like conformation, the X-ray structure of the Ras-like mutant of Rap1A, called Raps [Rap(E30D,K31E)], complexed with Mg 2+ GppNHp and Raf-RBD [7] can serve as a model. The most important interactions derived from the X-ray structure are depicted in Fig. 6. It is assumed to represent state 2 of the protein. Interactions assumed to be absent in state 2 and or weakened (or abolished) by the replacement of an oxygen atom with a sulfur Table 4. Affinities of Raf-RBD to Ras complexed with different GTP analogues. The association constant K A between Raf-RBD and Ras complexed with different GTP analogues was determined using ITC. Measurements were performed at 298 K in 40 mm Hepes NaOH ph 7.4, 10 mm MgCl 2, 150 mm NaCl, 2 mm 1,4-dithioerythritol. Data were analysed using ORIGIN FOR ITC 2.9 assuming a 1 : 1 complex formation [28] and DG ¼ G complex ) G free ¼ -RTlnK A. Raf-RBD complexed with K A (lm )1 ) DG (kjæmol )1 ) DH (kjæmol )1 ) TDS (kjæmol )1 ) Ras(wt) Mg 2+ GppNHp 2.50 ± 0.4 )36.5 ± 0.6 )13.4 ± ± 2.1 Ras(wt) Mg 2+ GppCH 2 p 2.50 ± 0.4 )36.5 ± 0.6 )18.4 ± ± 2.6 Ras(wt) Mg 2+ GTPcS 2.44 ± 0.6 )36.4 ± 0.9 )7.5 ± ± 2.4 Ras(T35S) Mg 2+ GppNHp 0.12 ± 0.04 )29.0 ± 0.06 )9.7 ± ± 1.1 Ras(T35S) Mg 2+ GppCH 2 p 0.09 ± 0.04 )28.2 ± 0.06 )15.3 ± ± 1.6 Ras(T35S) Mg 2+ GTPcS 0.18 ± 0.04 )30.0 ± 0.06 )13.6 ± ± FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS

9 M. Spoerner et al. Conformational dynamics of Ras bound to GTPcS A B C Fig. 6. Schematic representation of the coordination sphere of the phosphate groups and the thiophosphate of GTPcS in wild-type and mutant Ras nucleotide complexes. G, guanosine. (A) Coordination that predominantly exists in wild-type protein containing Thr35. (B,C) Other possible complexes with Ras(T35S) or Ras(T35A). Note, that not all contacts between the nucleotide and the protein are included. Bonds that probably exist only in state 1 or are weakened or abolished in the thiophosphate group are represented by broken lines. The sulfur atom was assumed to be negatively charged as shown previously for free ATPcS [32]. However, in the protein bound nucleotide the charge distribution is probably also influenced by the protein environment and could be thus different in different conformations. atom in the c-phosphate group are represented by broken lines. Influence of the nucleotide bound on the Ras conformational states 31 P NMR spectroscopy allows us to probe the conformational states of nucleotide-binding proteins, such as Ras-related proteins, which lead to structural rearrangement in the active centre. In principle, whenever chemical shift changes are visible they indicate that there is a change of the environment of the phosphorus nuclei, although small changes in structure can lead to large differences in chemical shifts and vice versa. The main mechanisms leading to changes in chemical shifts are conformational strain and electric field effects polarizing the oxygen atoms of the phosphate groups. In addition to these direct effects, long-range effects may occur, caused by a structuredependent change in the anisotropy of the magnetic susceptibility, with ring current effects making the most dominant contribution. Binding of the different GTP analogues to Ras leads to large changes in chemical shift, namely a strong upfield shift in the a-phosphate resonance and a strong downfield shift in the b-phosphate resonance compared with data from free Mg 2+ nucleotide complexes (Table 2). In complexes with GTPcS, a relatively small upfield shift of 0.63 p.p.m. is observed for the a-phosphate resonance and a strong downfield shift of 3.84 p.p.m. is observed for b-phosphate resonance. c-phosphorus resonances do not show the typical shift changes common to all analogues. Thus, qualitatively the phosphorus of the a-phosphate group in the magnesium complexes of GTP and its analogues is less shielded when bound to the protein, whereas the strong downfield shift in the resonance most probably results from strong polarization of the phosphorus oxygen bonds in the b-phosphate group. Such bond polarization in Ras Mg 2+ GppNHp has been discussed by Allin et al. [17], as an explanation of strong infrared shifts seen in the P O vibrational bands after complexation. It should be mentioned that the degree of shift differences in the chemical shift values cannot be related in a simple way to the degree of conformational change causing this change. Whereas wild-type Ras complexes with the GTP analogues GppNHp or GppCH 2 p exist in a conformational equilibrium between two main conformational states 1 and 2, with a K 12 value of 2, the complex with the analogue GTPcS obviously exists in predominantly only one conformation. It shows the spectral characteristics of state 2 as the effector binding state. (a) The interaction with Ras-binding domains leads FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS 1427

10 Conformational dynamics of Ras bound to GTPcS M. Spoerner et al. only to small chemical shift changes. (b) Weakening or destruction of the naturally occurring hydrogenbond interaction of the side-chain hydroxyl group of Thr35 with the metal ion, and of the main-chain amide with the c-phosphate by mutations to serine or alanine leads to large changes in chemical shift. (c) These chemical shift changes can usually be reversed in Ras(T35S) by Raf-binding because serine still contains a side-chain hydroxyl, however this is not the case in Ras(T35A). Geyer et al. [1] suggested that in the GTP-bound form, Ras(wt) also exists predominantly in one conformation. In terms of the conformational equilibria of Ras, GTPcS seems to be the analogue which is more similar to physiological GTP than both other commonly used analogues GppNHp or GppCH 2 p. Structural states of Ras(T35S) and Ras(T35A) Mutation of Thr35 to serine or alanine leads to two new phosphorus lines of the c-thiophosphate group and the b-phosphate group, which both show characteristics of state 1. The two states are in a dynamic equilibrium as evident from their temperature dependence. They are therefore assumed to represent substates of state 1 and are called states 1a and 1b. The alanine mutation makes coordination of the side chain with the divalent ion typical for state 2 impossible and can therefore only exist in state 1. In the serine mutant, metal ion coordination is perturbed but still possible. It shows, in addition to lines assigned to substates 1a and 1b, a very weak line at the position of the c-phosphate resonance in wild-type Ras, suggesting that Ras(T35S) shows in equilibrium a sparse population of state 2. As in the case of the complexes of Ras(T35A) or Ras(T35S) with the two analogues GppNHp and GppCH 2 p, the resonance of the a-phosphate is shifted downfield relatively to state 2, whereas the b-phosphate resonance is shifted upfield and is split into two. The c-phosphate resonance is also split into two well-separated lines, but one is shifted downfield and one upfield from the resonance positions obtained with the wild-type protein. As observed earlier for GppNHp and GppCH 2 p complexes of Ras, and now for GTPcS, not only is the hydroxyl group of Thr35 that interacts in the X-ray structures with the metal ion important for stabilization of state 2, but so too is its methyl group. This is evident because in Ras(T35S) an hydroxyl group remains available but state 2 is destabilized. Stabilization of state 2 by the side-chain methyl group of Thr35 does not seem to be due to a simple hydrophobic interaction, but rather to sterical restraints, because it is located in a cavity formed by the side chains of Ile36 and the charged polar side chains of Asp38, Asp57 and Thr58. In GTPcS bound to Ras three different stereoisomers of the thiophosphate group are possible (Fig. 6). In principle, they can occur in state 1 and state 2 of the protein, but the corresponding populations may differ greatly. However, they are not equivalent energetically because sulfur is coordinated more weakly to magnesium ions than oxygen and is a weaker acceptor of hydrogen bonds than oxygen. As a consequence, GTPcS binds more weakly to Ras than does GTP itself [18]. In state 2, the amide group of Thr35 is probably involved in a hydrogen bond with one of the nonbridging c-phosphate oxygen atoms and the divalent ion with the other oxygen; the third oxygen is probably involved in a hydrogen bond with the amide of Gly60 and the interaction with the positively charged side chain of Lys16. Energetically, a sterical position such as that shown in Fig. 6A is strongly favoured, in agreement with the experimental observation of a single phosphorus resonance for the c-phosphate (Fig. 6A). In the mutant proteins, state 1 is strongly preferred because the side-chain interaction of Thr35 with the Mg 2+ ion is perturbed (T35S) or impossible (T35A). It has been suggested previously [2] that weakening of metal ion coordination most probably leads to a concerted breaking of the hydrogen bond between the amide group of amino acid 35 and the c-phosphate group. Indeed, M-Ras [11] and H-Ras(G60A) [12] in the GppNHp form show 31 P NMR spectra typical of state 1 and recently published X-ray structures show that the amide group of Thr35 is distant from the c-phosphate group. Ford et al. [12] proposed a third conformational state for human wild-type H-Ras because their spectrum contained three 31 P resonances corresponding to the a- and c-phosphate (note that a new resonance assignment published by Spoerner et al. [2] was not known to Ford et al. [12]). However, because the third state could not be observed in our experiments, and the chemical shifts are very close to those observed for H-Ras Mg 2+ GDP, they should most probably be assigned to the a- and b-resonances of Ras-bound GDP. When a hydrogen bond exists between the amide group of amino acid 35 and the c-phosphate in the mutant proteins, in the GTPcS-complex the free energy differences DG and thus the equilibrium populations of the three stereoisomers are changed (Fig. 6). In the stereoisomer that most probably dominates in wild-type Ras (Fig. 6A), coordination of the c-phosphate group with the metal ion and the interaction 1428 FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS

11 M. Spoerner et al. Conformational dynamics of Ras bound to GTPcS with Gly60 and Lys16 is still possible, in the two other cases either the metal ion coordination is weakened when the sulfur is oriented towards the metal ion (Fig. 6B) or the hydrogen-bond interaction with Gly60 is weakened (Fig. 6C). It seems plausible that a wildtype-like arrangement is the energetically most favoured; meaning that state 1a should be assigned to this stereoisomer. For the resonances assigned to state 1b it is more difficult to derive a structural hypothesis. However, the chemical shifts of the c-phosphate resonance in state 1b are close to those observed in metal-free GTPcS (Tables 1 and 2) suggesting that it represents the arrangement seen in Fig. 6B, with coordination with the metal ion abolished. Dynamics and energetics of the conformational transitions The DG values for the transition between state 1 and 2 in complexes of wild-type protein with GppNHp and GppCH 2 p are 42 and 41 kjæmol )1, respectively [1,2]. For the complex between wild-type protein and GTPcS the activation energy cannot be determined by NMR because only state 2 is visible. However, it is reasonable to assume that the activation energy is similar. For the transition between states 1a and 1b we determined a DG value of 41 kjæmol )1. Thus the activation energies are identical within the limits of error. This may be due to chance or may suggest that a similar transition state is involved in the transition between states 1 and 2 and between states 1a and 1b. Dynamic equilibria in Ras complexed with GTP and different GTP analogues have been described previously [4,19,20] in 15 N-enriched Ras. It was shown that a number of amide resonances are not visible in 2D heteronuclear NMR spectra, most probably because of exchange broadening. In the complex between Ras(1 171) and GppNHp the amide resonances of 22 nonproline residues are not visible, whereas in the complex with GTPcS or GTP 20 additional resonances can be detected. Some of these resonances are broader in the GTPcS complex than in the GTP complex [4]. It is clear that any protein exists in multiple conformational states (now often called excited states) with different populations. Ito et al. [4] called this phenomenon regional polysterism. Different probes are also differentially sensitive to different conformational states. The two main conformational states 1 and 2 coexist with almost equal populations in the GppNHp complex, and exchange between theses two states, which probably involves structural changes in loop L1, switch 1 and switch 2, can qualitatively explain the excessive line broadening of the resonances of residues located in these regions. Additional local conformational changes may strengthen this effect. GTP and GTPcS exist predominantly in state 2 meaning that the line broadening associated with the transition will be smaller. The equilibrium between different stereoisomers around the thiophosphate group may contribute to the increased line width seen in some resonances in the GTPcS complex. Affinity of Ras-binding domains of Raf-kinase to Ras complexed with different GTP analogues ITC measurements show that under our experimental conditions the affinities between Ras(wt) and the tightly binding Raf-RBD are not influenced much by the type of bound GTP analogue. The association constant is identical within the limits of error for all three GTP analogues (Table 4). At first this seems surprising because NMR spectroscopy shows that the bound analogue clearly influences the equilibrium between the two conformational states. However, NMR spectroscopy indicates that, after binding of the RBD, Ras most probably exists in its correct structural state. Because the data analysis used assume a two-state model and the free enthalpy difference between state 2 in the free and complexed form can be assumed to be very small, only the conformational equilibrium with state 1 could influence the total enthalpy change measured by ITC. However, in the GppNHp or GppCH 2 p complexes of the wild-type protein, the difference in DG 12 between the two states is 2kJÆmol )1 (Table 3) and thus much smaller than DG involved in effector binding, which is of the order of 36 kjæmol )1 (Table 4). In addition, only the relatively small fraction of the protein occurring in state 1 in GTPcS would contribute to a nucleotide-specific variation in DG and thus would be scaled down proportionally to K 12 )1. In the T35S-mutant the population of free Ras is shifted to state 1, but the binding of Raf-RBD restores the correct conformation (similar to state 2). The wildtype and mutant proteins differ mainly in the methyl group of threonine which is missing in Ras(T35S). From the NMR point of view, Ras(wt) and the serine mutant seem to exist in the same conformation when bound to effectors. This is not true for the complexes of Raf-RBD with Ras(T35A) where the interaction with the RBD cannot restore the correct conformation [9]. For the T35S-mutant the dissociation constant increases by about one order of magnitude with the largest increase seen for GppCH 2 p. DG increases by 7.5, 8.3, and 6.5 kjæmol )1 for GppNHp, GppCH 2 p and GTPcS, respectively. The small differences may reflect FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS 1429

12 Conformational dynamics of Ras bound to GTPcS M. Spoerner et al. the energy difference between states 1 and 2 in this mutant, which cannot be derived from our NMR data. Relatively small, but in some cases significant, differences in DH and TDS values are seen in the three nucleotide complexes. Qualitatively, the differences may be rationalized with the help of the NMR results as follows. In a dynamic equilibrium Ras in complex with GppNHp or GppCH 2 p has a mobile effector loop which is fixed upon RBD binding. Therefore, the change in the configurational entropy (as part of the total entropy) is smaller than in the GTPcS complex, where the effector loop is suggested to be oriented in the correct position already. The nucleotide analogue bound to Ras influences the equilibrium between states 1 and 2. Replacing the oxygen bridging the b- and c-phosphate group in GTPcS with an imido or methylene group shifts the population of state 2 to state 1. Although it was shown that P O P bonds have very open bond angels [21], which should lead to delocalization of the electron density into the neighbouring atoms, the bridging oxygen may still be a weak hydrogen bond acceptor. That is an interaction involving a hydrogen bond donator and or a group with positive partial charge could be a reason for stabilization of state 2 in the GTPcS complex that is abolished by replacement of the bridging oxygen. Taking a closer look at the crystal structure of Ras in the GTP-bound state [22], only the main-chain NH of Gly13 and or the amino group of Lys16 and the bridging P b O P c oxygen seem to be able to contact each other by forming a hydrogen bond. The interacting groups are also close enough when GppNHp is bound, as derived from X-ray structure [23], although in this case a strong hydrogen bond is not to be expected. Conclusions Ras bound to triphosphate nucleosides exists in (at least) two conformational states which can be identified using 31 P NMR spectroscopy. One of these states (state 2) represents the high-affinity binding state for effectors; the second state (state 1) represents a different state of the protein with much reduced affinity to effectors. The equilibrium between the states can be shifted by using different GTP analogues or by specific mutations of Ras. A hydrogen bond of the amide group of Gly13 and or the amino group of Lys16 with the b c-phosphorus-bridging oxygen may be one factor responsible for stabilization of state 2 in the GTP complex. Thus, Ras(wt) Mg 2+ GTPcS exists predominantly in state 2. Other factors stabilizing state 2 are clearly the interactions of the amide and side-chain hydroxyl groups of Thr35 with the c-phosphate group and metal ion, respectively. Ras variants existing in state 1 show two substates states 1a and 1b. The transition velocity between these two states and thus the energy of the transition state is similar to that found for transition between states 1 and 2 of Ras bound to the analogues GppCH 2 p or GppHNp. The activation barrier may reflect a transient breakage of the bond between the metal ion and the c-phosphate. Experimental procedures Protein purification Wild-type and Thr35 mutants of human H-Ras(1 189) were expressed in Escherichia coli strain CK600K with ptac vector plasmids and purified as described previously [18]. Nucleotide exchange to GppNHp, GppCH 2 p or GTPcS was done using alkaline phosphatase treatment in the presence of a twofold excess of the GTP analogue as described at John et al. [24]. Free nucleotides and phosphates were removed by gel filtration. The final purity of the protein was > 95% as judged from the SDS PAGE. The Ras-binding domain of human craf-1 (Raf-RBD, amino acids ) was expressed in E. coli and purified as described previously [25]. Sample preparation Typically 1 mm Ras Mg 2+ GTPcS was dissolved in 40 mm Hepes NaOH ph 7.4, 10 mm MgCl 2, 150 mm NaCl, 2 mm 1,4-dithioerythritol and 0.1 mm 2,2-dimethyl-2-silapentane- 5-sulfonate in 5% D 2 O, 95% H 2 O. For binding studies a solution of 5 or 7 mm Raf-RBD contained in the same buffer was added in appropriate amounts to the samples. NMR spectroscopy 31 P NMR spectra were recorded with an Avance-500 NMR spectrometer (Bruker Biospin, Karlsruhe, Germany) operating at a 31 P frequency of 202 MHz. Measurements were performed in a 10 mm probe using 8 mm Shigemi (Tokyo, Japan) sample tubes at various temperatures. Seventy-degree pulses were used together with a total repetition time of 7 s. Protons were decoupled during data acquisition by a GARP sequence [26] with strength of the B 1 -field of 980 Hz. For referencing a X-value of reported by Maurer and Kalbitzer [27] was used, which corresponds to 85% external phosphoric acid contained in a spherical bulb. The assignment of the phosphate resonances was established by a 31 P 31 P NOESY experiment on 1.2 mm solution of Ras(wt) Mg 2+ GTPcS at 283 K with a mixing time of 1.5 s and a total repetition time of 14 s. Saturation experiments were performed at 278 K using a B 1 -field of 18 Hz for a period of 1 s for presaturation FEBS Journal 274 (2007) ª 2007 The Authors Journal compilation ª 2007 FEBS